I am getting 429 errors with message too many requests even when way under the requests per minute and token limits. I’m getting limited based on the number of requests, but I am only sending about 30 requests per minute, with no requests more frequent than every 2 seconds. I have verified the number of actual requests and tokens being sent both in my code and on the openai usage page. This is using gpt-4 in a paid account, so the limit should be 200 requests per minute I believe, more than 10x beyond what I’m getting limited at. Any ideas? Thanks for your help.
Rate limits are applicable on org. You’ll hit rate limit errors if another account is consuming the API in the same time frame.
You can check the rate limits applicable for your org here
There are two rate limits.
1: rate limit for the requests that YOU send
2: rate limit for the sum of everyone’s requests.
We’re all seeing the effects of 2, but the messaging is so confusing that we may think that it’s 1.
So, not your fault, we cannot do anything but wait for either everyone else to stop sending requests, or, for OpenAI to upgrade their servers.